Buried in the paperwork for one of the largest technology listings on record is a sentence that changes what a shareholder owns. Anthropic is going public as a public benefit corporation, a legal form that requires its directors to weigh something other than the share price.
The structure is not decorative. Under Delaware law, a public benefit corporation must be managed in a way that balances three things: the financial interests of stockholders, the best interests of those materially affected by the company’s conduct, and the specific public benefit named in its charter. Directors are protected when they weigh the third against the first. They are not required to rank returns last, but they are permitted to.
Anthropic’s version of that obligation arrives with an unusual governance layer. A long-term benefit trust, rather than the public shareholders, holds the right to elect a majority of the board, an arrangement designed to keep the company’s stated mission from being outvoted. The founder side of the ledger retains influence that ordinary investors in the offering will not share.
Public benefit corporations are legal in most states and rare on exchanges. The ones that have listed — Lemonade, Vital Farms, Warby Parker among them — are consumer and food businesses that went public at sizes where a governance discount was affordable. None carried a valuation within an order of magnitude of this one.
CNBC reported on Oct. 28 that the structure adds risk for public market investors, a dry sentence that covers a real problem. An index fund buying the shares cannot instruct the board to maximize cash flow, and an activist cannot win a proxy fight on the argument that a product decision costs margin. The charter converts a set of decisions that normally belong to shareholders into decisions that belong to directors with a second constituency.
The same day, Michael Burry used the filing to make a broader argument. The investor, who ran Scion Asset Management and became famous for betting against subprime mortgages, wrote in a public post that the prospectus shows an industry that is fragile at its core. His complaint was arithmetic rather than moral: commitments to buy computing capacity are far larger than the revenue those commitments are meant to produce, and the pace of fundraising does not match the time it takes to earn the money back.
Burry has made versions of this case before, arguing that the largest cloud operators understate depreciation on servers and chips and thereby overstate profit. The argument has been wrong on timing for longer than most short positions survive. It has also been directionally right about how capital-intensive businesses fail slowly, which is the reason it keeps finding an audience each time a filing discloses the size of the obligations.
The valuation question is separate and, for the underwriters, more urgent. Aswath Damodaran, a finance professor at New York University whose valuations of technology companies are widely circulated, put the company at $1.48 trillion, below the range that had been circulating before the filing. His figure is not a forecast of the first trade. It is a statement about what the cash flows would have to become to justify the price.
Those two arguments pull in opposite directions, and that is the awkward part of this listing. A charter that allows a board to accept lower returns in pursuit of a mission asks investors for patience. A valuation near the top of the market’s range asks them for growth at a pace that leaves little room for patience. The company is selling both propositions in the same document.
There is a defensible version of the combination. If the mission is genuinely an input to the product — a lab that can decline a contract, slow a release or restrict a model on safety grounds, and can say so in writing — then the structure is a form of product assurance that customers pay for. Enterprise buyers have started asking vendors what their governance permits. A charter with teeth is an answer.
The less flattering version is that the charter functions as insurance against a bad quarter. A board that may weigh public benefit can also cite it, and a shareholder who disagrees has limited recourse: the statutes allow a claim only in narrow circumstances, and the benefit report that accompanies the obligation is filed on a two-year cycle rather than with quarterly results.
Investors have seen dual-class shares, Up-C structures and founder-controlled trusts before, and the market has learned to price them by discounting the shares rather than arguing about the theory. What makes this case different is that the discount has to be applied to a company whose entire pitch is scale, and scale is the argument that governance restraints do not bind.
The first real test comes at pricing, when the book tells you whether institutions treated the charter as a footnote or as a risk factor. If the deal clears near the range the bankers want, the structure will be described as visionary by the same analysts who ignored it during the roadshow. If it prices below, the charter will be the explanation offered, and the argument about whether a mission can survive a public shareholder base will begin in earnest.
For the people running the company, the harder test comes later, in the first decision where the two duties diverge and the choice becomes public. That is the moment the paperwork stops being paperwork.
